I've been faced with the following generalized problem in a programming problem.
Give a $N$ faced die (each face has an unique number in the range $[1, N]$), what's the probability of getting a value greater than or equal to $S$, after rolling the die $M$ times and summing all the obtained values?
$N, S, M$ are all given as input.
I've managed to solve this through a brute-force solution, by counting combinations that had a sum $\geq S$, but I'm more interested in an efficient mathematical approach. I've seen this post Probability of dice sum just greater than 100, but it didn't help me to understand how I should proceed for this particular problem.
Generating functions and coefficient identification seem to be too expensive regarding time complexity.
So, what would be an efficient mathematical algorithm for this?